wilmott.com

What do you think of the TOTEM contribution for Equity correlation ?It is based on an historical evaluation. What I calibrated is that the contributors shift the historical correlation (log returns 5 days, deepness of the historical data : 3 to 5 years depending the bank) with a baricentric choc to value 1 of 17%.Any comments on this subject ? What about a term structure ? Other ways to calibrate a geographical or sectorial implied correlation as it is done on the CDOs markets ?Best Regards

I have tried to use TOTEM for correlation between FX long-dated vols. This data is monthly and history is not very long (a few years). It is said that banks tend to "cheat" on valuable data but it gave reasonnable estimates for vol of vol. However I stumbled on the correlations. I suspect that is because of the sparsness : netting the variations over a month makes you loose information about simultaneity of changes somehow. The correlation I get seems too low when I compare it to the proxy I get from better data on short-dated vols, extended with some sort of one factor HJM on the term structure.Do you really need TOTEM for something a liquid as equity spots ?